Page 213 - Anatomy of a Robot

08_200256_CH08/Bergren 4/10/03 4:39 PM Page 198
198 CHAPTER EIGHT
A/D Conversion
We’re not going to discuss the types of A/D converters that are available, nor are we
going to discuss how they work. We leave it up to the reader to delve into these details,
including cost and linearity. Just remember that it must be fast enough to keep up with
the sample rate chosen according to the Sampling Theorem. Here are a few good URLs
that talk about A/D conversion in general:

http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/adc.html
http://jever.phys.ualberta.ca/ gingrich/phys395/notes/node151.html
www.sxlist.com/techref/io/atod.htm

We do need to have a discussion about the number of bits in the A/D. First of all, we
must recognize that an A/D converter’s primary characteristic tends to be the number
of bits in the digital output. Be wary of A/Ds that have many bits. It’s not unusual for
an A/D to fail to perform up to its reported level. So even if an A/D touts 16 bits of res-
olution, it may only deliver the equivalent performance of 12 or 14 bits. It seems obvi-
ous that a real-world signal cannot be well represented by just 2 or 3 bits of data. But
how many bits do we really need?
First, we need to define db or decibel. This acronym has many uses, which each have
their own definition. Here we will take it to mean a method of measuring voltage ratios.
A voltage signal that is 6 db lower than another is just 50 percent of the other. Increasing
a voltage signal by 6 db doubles it. In a similar manner, 20 db connotes a factor of 10.
A good web site on decibels is at www.its.bldrdoc.gov/fs-1037/dir-010/_1468.htm.
The primary consideration when looking at A/D bit length is the nature of the input
signals. What signal-to-noise (S/N) ratio do the signals have? All signals have noise on
top of them. Some signals have far more than others. If a signal is roughly 10 times big-
ger than the noise, then it is 20 db S/N. Figure 8-5 shows a visual representation of noise
at different S/N ratios.
It’s important to know the S/N ratio of the signals being measured. The rule of thumb
is that each extra bit in the A/D provides another 5 db of S/N capability in the DSP
engine. Ordinarily, another bit would double the effective range of a word and thus pro-
vide 6 db of S/N capability, but I’ve been told by experts not to expect the theoretical
limit, so count on 5 db per bit.
Now if the signal to be measured has a 40 db S/N ratio, then an 8-bit A/D might be
just the ticket since 8 5 40. As long as stepping up to a couple of more bits is not
too expensive, I’d consider a 10-bit A/D for such a job. Buying a 16-bit A/D will not
convey any extra accuracy with such a low S/N signal. Ordinarily, a 16-bit A/D might
allow 80 db of S/N processing (5 16), but if the input signals are not up to that num-